Surgical safety of cervical pedicle screw placement with computer navigation system

The radiological accuracy of CPS placement was evaluated using postoperative CT scans and was defined retrospectively by a grading scale originally proposed by Neo et al. [20]. (Fig. 3). Screw positions were classified into four grades: grade 0, no perforation, and the screw was completely contained in the pedicle; grade 1, perforation <2 mm (that is, less than half of the screw diameter); grade 2, perforations ≥2 mm but <4 mm; and grade 3, perforation ≥4 mm (complete perforation). A total of 452 CPSs were placed using a preoperative 3D CT-based navigation system (first period of 2007–2012), and a total of 310 CPSs using an intraoperative 3D CT-based navigation system (second period of 2012–2015), so that 762 CPSs were placed in the spine level of C2 to Th3. In the first period of 2007–2012, accuracy of CPS placement was 93.6 % (423 of the 452 screws) in grade 0, 3.1 % (14 of the 452 screws) in grade 1, 2.7 % (12 of the 452 screws) in grade 2, and 0.6 % (3 of the 452 screws) in grade 3. In the second period of 2012–2015, accuracy of CPS placement was 97.1 % (301 of the 310 screws) in grade 0, 2.3 % (7 of the 310 screws) in grade 1, 0.3 % (1 of the 310 screws) in grade 2, and 0.3 % (1 of the 310 screws) in grade 3. Accuracy of CPS placement was also analyzed based on spine level. Screw malposition was found 0.9 % (1 of 114 screws) at C2, 5.5 % (5 of 91 screws) at C3, 5.9 % (8 of 135 screws) at C4, 4.9 % (6 of 122 screws) at C5, 3.3 % (3 of 88 screws) at C6, 10.2 % (10 of 98 screws) at C7, 5.9 % (4 of 68 screws) at T1, 2.9 % (1 of 34 screws) at T2, and 0 % (0 of 12 screws) at T3. The overall rate of screw malposition rate at C2-C6 (where VAs are usually running in the transverse foramen) was 4.1 % (23 of 550 screws). Illustrative case of posterior cervical instrumented fixation using CPS was presented in Figs. 4 and 5

In total, 38 of the 762 screws (5.0 %) were found to perforate the cortex of pedicle in the first and second period, although any neural or vascular complications closely associated with CPS placement were not encountered in the present study. There was no case of vertebral artery injury related to CPS placement. Twenty nine of the 38 screws (76.3 %) were found to perforate laterally, 7 screws (18.4 %) were found to perforate medially, and 2 screws (5.3 %) were found to perforate inferiorly. There were no screws that were found to perforate superiorly. Accuracy of CPS placement was summarized in Table 2.

In the present study, we focused our attention on the safe CPS placement for posterior cervical instrumented fixation. Careful imaging analysis demonstrated that a total of 38 of the 762 screws (5.0 %) were found to perforate the cortex of pedicle in the first and second period, although any neural or vascular complications closely associated with CPS placement were not encountered. CPS misplacement was more recognized in C7 compared to the other spine level, and the misplacement to lateral direction tended to occur in mid-lower cervical spine. CPS at C7 is usually applied in case of posterior cervical long fusion, and placed at the caudal side of fusion segment. The procedure of CPS placement at C7 level may be easily affected by the inward pressure of well-developed paravertebral muscle layers. The stronger the axial trajectory is kept against the medial pressure from the paravertebral muscles, the easier the axial rotation or sagittal bending of the cervical spine might be, so that screws may be misplaced laterally. The accuracy of CPS placement may be also affected by anatomical complexity of cervical pedicle or navigation error.

In 1991, Panjabi et al. published the first three-dimensional anatomical study of the human cervical vertebrae [21]. The three-dimensional coordinates of various marked points on the surface of the vertebra were measured with a specially designed morphometer instrument. Transverse pedicle diameter increased from 5.1 mm at C3 to 6.6 mm at C7, and sagittal diameter increased from 6.7 to 7.6 mm across the same levels. Karaikovic et al. investigated cervical pedicle morphology using human cadaver specimens via manual and computed tomography measurements and suggested that there were anatomical limitations associated with the use of the pedicle screw in the cervical spine [22, 23]. Chazono et al. evaluated the linear and angular parameters of the cervical spine as well as pedicle dimensions required for safe CPS placement using multiplanar CT reconstruction [24]. They reported that the overall mean pedicle transverse angle (PTA) ranged from 33.6 to 50.2° and that the smallest mean PTA was found at C7 in males (33.4°) and in females (34.1°); the largest mean PTA was found at C4 in males (50.4°) and in females (49.6°) followed by C5, C3, and C6. Further, the pedicle sagittal angle (PSA) varied from 13.7° in the cephalad direction at C3, 5.3° at C4, 1.3° caudally at C5, 3.0° caudally at C6, and 3.3° caudally at C7. The suggested that not only pedicle dimensions but also linear and angular parameters of the cervical vertebral body can be useful data for safe CPS placement. Wang et al. demonstrate their technique of free-hand subaxial CPS placement without using intraoperative navigation devices, and reported clinical results in which C3 showed a higher perforation rate than the other vertebrae [25]. They analyzed the transverse angle of CPS trajectory and perforation of the lateral wall by using postoperative CT scans, and suggested that there are two crucial maneuvers for increasing accuracy of screw placement: identifying the precise entry point and guiding CPS trajectory on axial plane by the resistant of thick medial wall. Shin et al. reported that the pedicle size and angle as well as the shapes of the cervical pedicle are important for safe CPS insertion [26]. Twenty-six human cervical vertebrae from fresh-frozen spines were secured to a thin sectioning apparatus to produce three 0.7-mm-thick pedicle slices along its axis. Radiographs taken of these pedicle slices were scanned, digitized, and traced to facilitate visual comparison. The pedicle slices were found to exhibit substantial variability in composition and shape, not only between individual spines and vertebral levels but also within the pedicle axis. They suggested that the lateral cortex was consistently found to be thinner than the medial cortex in all samples. They reported that pedicle axial slices in the mid-cervical vertebra exhibit a semi-circular or rectangular or triangular shape, with the flat surface directed towards the lateral aspect of each pedicle. As the slices progress antero-posteriorly along the pedicle long axis, the shape of the pedicle slices changes to a more circular conformation. Surprisingly, there are some slices that exhibited an unusual hook shape at their inferior aspects. These anatomical findings must be noted by surgeons attempting CPS placement for posterior cervical instrumented fixation.

The percentage of CPS misplacement varies from 2.5 to 29.1 % in the literature (Table 3) [3, 16, 20, 27‐36]. In the present study, 6.4 % of CPS misplacement of grade 1–3 in the first period and 2.9 % in the second period were recognized. Intraoperative image guidance is an important issue to avoid the surgery-related complications. Abumi et al. reported that 45 of 669 inserted screws (6.7 %) were misplaced in his early series [3]. Neo et al. reported that the rate of malposition was 29 % in patients with degenerative conditions [20]. Hojo et al. reported that the overall rate of malposition was 158 of 1065 inserted screws (14.8 %) in their retrospective multicenter study of free-hand insertion technique with fluoroscopy [16]. Even with a preoperative 3D CT-based navigation system, there is always a risk of misplacement of CPS that may result in significant neural or vascular complications. Intervertebral anatomical relationships with the patient in the prone position during surgery might not match the preoperative CT data obtained while the patient was in the supine position. This discrepancy has led to navigation errors. Intraoperative 3D CT-based navigation system may be the solution for navigation errors. It can reduce the discrepancy so that it may provide greater accuracy and feasibility during CPS placement. Ito et al. reported a malpositioning rate of 2.8 % (5 of 176 screws) for CPS placement using a 3D fluoroscopy-based navigation system (Iso-C^3DC-arm, Siemens Medical Solutions, Erlangen, Germany) [29]. Ishikawa et al. also reported that a 3D fluoroscopy-based navigation system can improve the accuracy of CPS when compared with a conventional insertion technique [27]. Intraoperative 3D CT-based navigation system is the cutting edge of image-guided surgery that includes the acquisition of high-resolution images with 3D data during the surgery and enables surgeons to use them as fully automatic navigation. Although the quality of O-arm images may not be fully comparable to the images of multidetector helical CT, particularly at the cervicothoracic junction which is crucial for CPS, surgeons may not need to rely on the preoperative CT images in which the patient’s position may vary from the surgical position during surgery.

CPS misplacement of grade 2 and grade 3 is clinically most critical. CPS misplacement of grade 2 and grade 3 in the present study was 3.3 % (15 of the 452 screws) using a preoperative 3D CT-based navigation system, and CPS misplacement of grade 2 and grade 3 was 0.6 % (2 of the 310 screws) using an intraoperative 3D CT-based navigation system. This advancement in image-guided surgery can improve the accuracy of CPS placement to secure the safe surgery. Lateral fluoroscopy in addition to computer navigation was also crucial and used for the accurate guidance of CPS placement, especially in case of high cervical or mid-cervical spine. However, in case of lower cervical or upper thoracic spine, CPS placement was basically guided only by using navigation, not lateral fluoroscopy, because the patient’s shoulder or scapula may disturb the surgeons to see the direction of CPS clearly. Additionally, several technical considerations without navigation system must be discussed. Abumi et al. originally demonstrated his original technique to create a funnel-like-shaped hole at the screw insertion point down to the entrance of the pedicle to secure a wider range of freedom for the screw insertion angle [2, 37]. Yukawa et al. used pedicle axis views generated by fluoroscopy during CPS placement [36]. Miyamoto et al. described a CT cutout technique to navigate the direction of the pedicle [32]. Recently, a patient-specific screw guide template system has been also introduced [30, 38].

The use of CPS placement for posterior instrumented fixation needs to be carefully determined based on the individual pathological and anatomical condition. A thorough knowledge of surgical anatomy, biomechanics, and technique remains the most essential aspect for successful surgery. Various attempts to avoid misplacement of CPS were performed but have not been completely successful. Therefore, we should continue to attempt to minimize the malposition rate by using a variety of techniques. Image-guided CPS placement has been an important advancement to avoid the surgery-related complications.